LA early voting initial results favor Republicans
Varying by degree only in whether early voting has matured within the electorate, Republicans initially have received good news from the practice this election cycle in Louisiana.
Now in its eighth year in the state, the last round of statewide elections topped the one-fifth mark for proportion of vote that came prior to election day. After so much experience, to understand how early voting may signal the direction of an election, we must consider whether it has come to approximate final returns, or whether it remains a phenomenon unrepresentative of actual outcomes.
For analytical purposes, partisanship and race constitute the most crucial variables. Party identification, for obvious reasons, remains a strong predictor of the vote, as does race, for historical and societal reasons. Thus, when the actual electorate participating in a given eleciton, for example, has a higher proportion of Republicans than reflected in registration statistics, GOP candidates gain an advantage. Similarly, for example, a higher proportion of blacks would bring disadvantage to that party’s candidates.
But such an approach ignores that these variables also stand in for other attributes that affect turnout. Most principally, because those who identify as Republicans typically score higher in terms of socioeconomic status – for example, are better educated than the typical Democrat – and blacks – again with education, in this case typically less than whites – score lower on SES, all things equal Republicans turn out at higher rates than Democrats, and blacks at lower rates than whites.
So, an adjusted normal vote provides a better comparison – the expectation that, all other things equal, Republicans will turn out disproportionately higher than Democrats, and whites disproportionately higher than blacks. Again, that holds constant all other factors – which sometimes doesn’t happen. For example, because of the reelection bid in 2012 by Democrat Pres. Barack Obama, who identifies as black, this mobilized blacks at an unusually high level. In Louisiana, holding constant partisanship and reviewing the label under which the large majority of blacks identify, in that election white Democrats turned out at a rate of 69.31 percent, but trailed black Democrats who voted at a rate of 70.94 percent.
If early voting has reached a state of maturation, the proportions witnessed in early turnout would essentially match those of all voting. That is, the typical early voter would not really differ in characteristics from the typical election day voter. If differences persist, then early voting is an act of political participation qualitatively different from voting on election day and represents a substitution effect – people of certain characteristics associated with partisanship and race are more likely to vote early.
The problem for analysts is, like Minerva’s Owl flying only at night, we don’t know whether maturation has been reached or a substitution effect persists until after the election itself when the data become available for actual turnout, meaning whether raw early voting numbers predict the actual proportions that lead to forecasting better or worse expectations for certain candidates and parties, or if these need adjusting to account for a substitution effect. To make the task more complex, a dynamic within early voting itself could exist where disproportionate turnout occurs among groups earlier and later in the process.
Taking all of the above into consideration, reviewing the first day’s results of early voting (note: link accurate only through Oct. 26, 2016) should lend optimism to Republicans in Louisiana. Registration and turnout data in Louisiana for the last presidential election in 2012 showed Democrats turned out about a percentage point higher than their proportion of the registered electorate, and Republicans at two-and-two-thirds higher than that. (Other party/no party registrants turn out significantly disproportionately lower levels, because disproportionately those who sign up this way have less interest in politics.) Between whites and blacks, whites turned out about 1.8 percentage points higher than their proportion among the registered, and blacks turned out about half that lower.
Comparing actual turnout to early voting in the 2012 election, Democrats turned out early at about one-and-three-quarters points proportionally higher than on election day while Republicans’ rate was about two-and-a-half times higher than that. (Other party/no party registrants turned out early significantly less proportionally.) As far as race goes, around 3 percentage points more blacks voted early than who would on election day, while for whites around 2.5 percentage points fewer voted early than who would on election day.
So if the same dynamics held this election, we should see similar kinds of numbers. We don’t, at least from the first day’s numbers: GOP early voting is about 3.5 points higher in 2016 than 2012, while Democrat early voting is almost the same lower; and for whites early voting is around 6 points higher, while black early voting is slightly more in the opposite direction.
If we assume the phenomenon has matured, white final turnout will be about 4 points higher in 2016 than 2012 and black turnout about the same lower. Under a similar assumption, Democrat final turnout will be almost two points higher, but swamped by an almost eight point increase for Republicans.
Continuing to assume a substitution effect, Republicans look still more advantaged. A 16.5 percent proportionate gap between Democrats and Republicans in 2012 early voting, which translated to a 19 percent advantage on election day, this year starts at only 10 percent which would predict just a difference of just 11.5 percent on election day. A 40.5 percentage point higher white proportion in 2012 early voting ended up as a gap of 36 on election day, but with the early voting initial difference having ballooned to 43 points this year, this portends a gap of over 38 percent on election day.
A relatively higher proportion of 7.5 percent Republican and 2 percent white on election day in 2016 as opposed to 2012 could spell some difficulties for down-ballot Democrat candidates. While GOP presidential nominee Donald Trump should triumph easily over Democrat nominee Hillary Clinton, a much more competitive Senate race could see two Republicans making an inevitable runoff instead of one candidate from each party. The same dynamic could play out in the Fourth Congressional District contest. For local races, the effect would magnify, as those kinds of contests typically have fewer defections from partisanship in making vote choices.
Should the first day results replicate across the entirety of the early voting period to its conclusion, Louisiana Democrats could find themselves in for a long night the week after.
In any given election, the proportion of individuals with certain characteristics in the electorate that end up voting can be to the advantage or disadvantage of parties and candidates. A crude method to determine this begins with the concept of a normal vote, or one where turnout essentially matches the characteristics of the aggregate of registered voters.
Posted by Jeff Sadow at 12:55